Relative Permittivity of Carbon Dioxide + Ethanol Mixtures prediction by means of Artificial Neural Networks

Abstract

CO2 + ethanol mixtures have a huge scientific interest and enormous relevance for many industrial processes. Obtaining of their chemical and physical properties is a fundamental task. Relative permittivity (r) of these mixtures is a key property because allows a better knowledge of the structure and the interactions in other media. In this work predictive values of relative permittivity (r) of carbon dioxide + ethanol mixtures were obtained implementing artificial neural networks (ANNs). They are used successfully in very different fields; therefore it is a very useful tool. In this case the obtained results enhance the ones from the usual multiple linear regression analysis. In both cases mass fraction, pressure and temperature experimental data from a direct capacitance method were used.